[Trilinos-Users] bug in EpetraExt_HDF5?

[Simone Deparis]

Simone Deparis wrote:
> Hi,
> 
> (trilinos 8.0.7)
> I suspect a bug in EpetraExt_HDF5.cpp, line 902:
>     Epetra_Map LinearMap(X.GlobalLength(), 0, Comm_);
> 
> if the roiginal multivector has indexBase == 1, the temporary
> multivector LinearX has one missing entry.
> 
> Can someone confirm, please? If I put
>     Epetra_Map LinearMap(X.GlobalLength(), X.Map().IndexBase(), Comm_);
> 
> should work?
> 
> Thank you
> best
> Simone

Thanks to Mike Heroux, the problem is fixed:
changes in EpetraExt_HDF5.cpp

902c902
<    Epetra_Map LinearMap(X.GlobalLength(), X.Map().IndexBase(), Comm_);
---
>    Epetra_Map LinearMap(X.GlobalLength(), 0, Comm_);

930c930
<   hsize_t offset[] = {0, LinearX->Map().GID(0)-X.Map().IndexBase()};
---
>   hsize_t offset[] = {0, LinearX->Map().GID(0)};


As far as I understood, this checnges are already in the development
branch. When will they be part of the distribution?

Thank you again
best
Simone